DSPy Signature -> Prompt -> Nano Banana

image_generation_from_prompt.py

import os
from google import genai
from google.genai import types
import dspy

# --- Configuration ---
DEFAULT_MODEL_NAME = 'gemini/gemini-2.0-flash'
IMAGE_MODEL_NAME = 'gemini-3-pro-image-preview'
# do create the directory
DEFAULT_SAVE_PATH = os.path.expanduser("~/image_generator/img/")

# --- Setup ---
def configure_dspy(model_name: str = DEFAULT_MODEL_NAME) -> None:
    """Configures the DSPy language model globally."""
    lm = dspy.LM(model_name)
    dspy.configure(lm=lm)

# --- Signatures ---
class OpArtSignature(dspy.Signature):
    """
    Generates an art prompt in the style of Op Art (Optical Art).
    Focus: Optical illusions, vibrating patterns, visual tricks.
    """
    visual_effect = dspy.InputField(desc="Visual effect: Vibrating colors, warping planes, motion, flashing")
    patterns = dspy.InputField(desc="Patterns: Repeating lines, concentric circles, moiré patterns, distorting grids")
    color_palette = dspy.InputField(desc="Color palette: High-contrast black and white, or intensely clashing bright colors")
    precision = dspy.InputField(desc="Precision: Meticulous, machine-like exactitude")
    image_generation_prompt = dspy.OutputField(desc="A detailed description suitable for generating an image in the Op Art style. Generated image shouldn't go in a frame or on a wall.")

class ImpossibleGeometrySignature(dspy.Signature):
    """
    Generates an art prompt in the style of Impossible Geometry (Optical Illusion).
    Focus: Visual paradoxes, spatial ambiguities, and objects that defy physics (e.g., M.C. Escher).
    """
    illusion_concept = dspy.InputField(desc="Concept: The specific paradox (e.g., Penrose triangle, infinite staircase, waterfall, tessellation)")
    perspective_trick = dspy.InputField(desc="Technique: Isometric projection, forced perspective, or dual-plane interpretation")
    forms = dspy.InputField(desc="Forms: Polyhedra, interlocking architectural elements, connected loops")
    rendering_style = dspy.InputField(desc="Rendering: Realistic shading and texture (stone, wood) to make the impossibility convincing")
    image_generation_prompt = dspy.OutputField(desc="A detailed description suitable for generating an image of impossible geometry.")

# --- Core Logic ---
class ImageGenerator:
    """Encapsulates image generation logic using Google GenAI."""

    def __init__(self, model_name: str = IMAGE_MODEL_NAME):
        self.client = genai.Client()
        self.model_name = model_name

    def generate(self, prompt: str, filename: str, save_path: str = DEFAULT_SAVE_PATH, 
                 aspect_ratio: str = "16:9", image_size: str = "4k", show: bool = True) -> None:
        """
        Generates an image from a prompt and saves it to the specified path.
        """
        print(f"Generating image for prompt: '{prompt}'...")
        try:
            response = self.client.models.generate_content(
                model=self.model_name,
                contents=prompt,
                config=types.GenerateContentConfig(
                    tools=[{"google_search": {}}],
                    image_config=types.ImageConfig(
                        aspect_ratio=aspect_ratio,
                        image_size=image_size
                    )
                )
            )

            image_parts = [part for part in response.parts if part.inline_data]

            if image_parts:
                image = image_parts[0].as_image()
                os.makedirs(save_path, exist_ok=True)
                
                full_path = os.path.join(save_path, filename)
                image.save(full_path)
                print(f"Image saved successfully to: {full_path}")
                if show:
                    image.show()
            else:
                print("Warning: No image data found in response.")
        
        except Exception as e:
            print(f"Error during image generation: {e}")

# wrapper for backward compatibility or simple usage
def generate_image(prompt, filename, save_path=DEFAULT_SAVE_PATH, aspect_ratio="16:9", image_size="4k", show=True):
    generator = ImageGenerator()
    generator.generate(prompt, filename, save_path, aspect_ratio, image_size, show)

def test_generate_image():
    """Test function for generating a simple infographic."""
    generate_image(
        "Generate an infographic of the current weather in Bengaluru.", 
        "weather_bengaluru.png", 
        "."
    )

# --- Main Execution ---
if __name__ == "__main__":
    configure_dspy()
    
    # Example usage with DSPy
    impossible_geometry_prog = dspy.ChainOfThought(ImpossibleGeometrySignature)
    
    # We pass only the relevant fields for Impossible Geometry
    prediction = impossible_geometry_prog(
        illusion_concept="infinite staircase",
        perspective_trick="dual-plane interpretation",
        forms="connected loops",
        rendering_style="blank and white lines"
    )
    
    prompt = prediction.image_generation_prompt
    print(f"Generated Prompt: {prompt}")
    
    # Generate the image
    generate_image(prompt, "impossible_geometry_test_4.png", ".")